Re: Am I being a bit thick here
"As of the end of 2014, the Home Office's CAID national database held 4.4 million images and over 30 million hashes."
Given the number of hashes exceeds the number of images then it would seem they are in some way dealing with possible duplicates.. The article info-graphic also mentions PhotoDNA hashes "to identify images even if the image has been altered".
"Dragon-Slayer"... OK it might be flawed or subject to avoidance by other means but then 'crime investigation' often relies on someone making a dumb mistake resulting in leads elsewhere.